A probabilistic framework for online test-time adaptation
This work provides a principled probabilistic approach to online test-time adaptation, which is important for deploying models in non-stationary environments.
The paper introduces a probabilistic framework for online test-time adaptation, addressing distributional shifts between training and test data. The framework uses state-space modeling to unify parameter learning, time evolution, prior tuning, and prediction.
This paper presents a probabilistic framework for online test-time adaptation problems. In them, a model is trained on labeled data but must adapt to unlabeled data at test time under the assumption that training and test distributions potentially differ, that is, there might have been a distributional shift. The framework is based on a state-space modelling architecture from which parameter learning, parameter time evolution, prior tuning, and prediction can be characterized.